Brandubh 7x7

Tafl rules
User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Sat Oct 03, 2020 4:17 pm

Copied from the small forum:

10-02/13:47 ChaolAibhne: @Hagbard, in my game as black against mmgaari at Brandubh 2 I've just been surprised to have the system impose a loss on me on the grounds of repetition. There had warnings about repetition, but it wasn't specified who was doing so; it was only after I had lost that a message said that it was black. I assumed that it was white that was being warned. My move 15 threatened an attack on one of white's pieces and his response threatened one of mine. My move 17 had no direct threatening aspect to it. Is it not white's move 18 that should be judged as the start of the repetition?

10-02/16:41 Adam: @ ChaolAibhne - after looking carefully, I agree with you, the computer is seeing a reapeat of a board position that you first set up on move 15, but what is not being repeated in your final move is 'how' that position came to be. So I would argue that the computer should be overruled in this instance. @Hagbard - from a programming perspective, could the computer be told to look for a board position which has been repeated in the same 'way', not just repeated visually? as in: IF 'board position' AND 'move that led to board position' are repeated 3 times end game. (in my VERY clumsy pseudo code). Or maybe all the pieces could be given an identity, such that a repeat of a single piece shuffling could be detected? This would also deal with detection problems when white shuffles as black does lots of different moves.

10-03/10:44 Hagbard: @ChaolAibhne: I've had a look at the game.
The comp detects that black attempts to break through from line E to line G by consistently moving a black piece e2-e3-e2 etc. (attacking), and a white piece consistently blocks the way for this black piece by moving f2-f3-f2 etc. (defending). After a number of repetitions of this pattern, the blocking (defending) side won.
But as you write, what is going on can also be seen in a different light: the white piece consistently attacks the black piece to be captured, and the black piece consistently evades this attack.
The comp detects for the first case (side stepping for break through), but not for the second case (side stepping for avoiding capture).
Interesting example!

10-03/15:55 Hagbard: Is there a repetition? Yes indeed! Who causes it? That is uncertain here, both could be causing it.
Is black moving back and forth to break through or to avoid being captured? It is not obvious that it would be an advantage to break through, but who knows the player's plan? The computer does not know the player's plan and neither would an umpire.
Is white moving back and forth to block black from breaking through or to capture the black piece? Also noone knows.
I suppose that an umpire should call this situation a repetition from both sides and a draw!

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Sat Oct 03, 2020 4:20 pm

unhandyandy wrote:
Sat Oct 03, 2020 2:50 am
I'm in favor of giving a win to Black in case of a repetition. That would be very clear and easy to program.
Of course, unhandyandy's suggestion to have black win repetitions would solve it all. It is practically always white who pressures to get out anyway.

Auskor
Posts: 8
Joined: Sat Sep 19, 2020 9:36 am

Re: Brandubh 7x7

Post by Auskor » Sat Oct 03, 2020 4:57 pm

So a very clear and straightforward repetition rule to use would be if the same BOARD POSITION is detected on 3 occasions then black is declared the winner irrespective of the sequence of moves that bring about the final position.

I don't know if it would have an effect on the game balance but from both players point of view it would be easy to understand. A system warning when the same position has been repeated twice would also be helpful.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Sat Oct 03, 2020 5:19 pm

Auskor wrote:
Sat Oct 03, 2020 4:57 pm
So a very clear and straightforward repetition rule to use would be if the same BOARD POSITION is detected on 3 occasions then black is declared the winner irrespective of the sequence of moves that bring about the final position.

I don't know if it would have an effect on the game balance but from both players point of view it would be easy to understand. A system warning when the same position has been repeated twice would also be helpful.
It presumably would not affect game balances, as repetition already typically is a white thing, and white lost from it.

Casshern
Posts: 18
Joined: Fri Jun 23, 2017 7:20 am

Re: Brandubh 7x7

Post by Casshern » Sun Oct 04, 2020 8:29 am

Please forgive my ignorance here, as I do not play Brandubh. Although move 15 is an attack on a white piece, white could move f3-g3 or king to d3. Not sure if either of these moves is strong or if white is simply in a losing position. The move repetition clearly starts on move 16 when white moves f3-f2 and black has to protect the piece while still trying blocking the corner (reacting to the aggressor).

I’m confused why the computer recognized this as a win for white. I remember a game that I lost as black on PTO a couple years ago, that clearly should not have been a win for white. It was a Copenhagen game and the situation was something like, black had a piece on b2 and white had the king on f3. The king moved to a3, forcing black to move to a2. Then, the king started to move back and forth between b3 and a3. In turn, black moved between b2 and a2 to block the king. The computer recognized this as a repetition win for white because the first move in the repetition sequence was not f3-a3 by the king, but b2-a1 by black. It makes sense to me why the computer came to this outcome. It is simply looking for a repetitive sequence of moves and whoever starts that sequence loses. I then understood that the computer cannot recognizes the aggressive player.

I have come to think of the repetition rule as a way for white to salvage a game in a losing position. This rule often comes into play in Sea Battle games. Black can set a trap or net for white to fall into by cutting him off from the rest of his men. In white’s desperation to escape the trap/net he might initiate a repetition. White is seen as the aggressor because if black does not block, white will reach an edge and win. But by stopping the repetition, white would allow black to close the net around the king and lose the game. This is very common in Sea Battle games.

Another example I could think of is, a Copenhagen game were the king got to a third row and black didn’t have any pieces on any third rows (3,9,c,i). The king could perpetually check every corner and would eventually lead to a repetition. In that case, white would have to get a piece on the third row to create a guillotine. At which time, black would also move a piece to the third row to block. Now, understanding that this line of attack leads to repetition, white should try to push for an advantage in a single corner (or edge for a fort), instead of meaninglessly threatening the corners with no way to win.

In my experience, white is always the aggressive player in these situations. There could be a situation where black is the aggressor, but I cannot think of one. I think it is a result of a losing position by white. In my experience, white usually resorts to a repetition because he has been out played my black. If white had any other options he would go for the them and the win. In chess, players repeat a position to create a stalemate/draw because neither player was able to create an advantage. In Tafl, this is not the case. A repetition usually occurs because black has created an advantage. Leaving white with no other option but to perpetually stale the game.

So, as many players have already said, I think the repetition rule should simply be a win for black, not a lose for the “aggressive player.” Black should not be punished with a lose (or draw like chess), when he has created a winning position/advantage. I hope this explanation also answers the questions/positions that Ytreza has proposed.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Mon Oct 05, 2020 2:50 pm

Further copied from the small forum:

10-03/18:33 Adam: @Hagbard, sorry, i disagree. I think its pretty clear. The rule says: Perpetual repetitions are forbidden. The player who causes the repetition must find an alternative move or else he loses the game. ChaolAibhnes move 15 creates the board pattern that ends up being the repetion that the computer spots, but it's not a repetition of ChaolAibhnes previous moves, as the computer is counting move 15, which isnt part of the repeated sequence. Even if white is reacting to move 15, white is starting the cycle of repeted movement here. Now, I think the rule ought to be spelled out more clearly, a repetition of movement , not a repetition of board pattern, which just happens to be how the computer spots a repetition, and in this case eroniously. A repetiton of board pattern is clearly not what the rule means. What if a board pattern was repeated that last happend 20 moves ago after much else has happened and pieces within the pattern have been exchanged? It would be absurd.

10-03/20:57 Hagbard: @Adam: I saw the replay again, and it is as you say. White is the first to attack with move 16, and black's move 17 would thus be an escape from the attack more than a break through attack of its own. Although in another game black could kill two birds with one stone and do an escape and attack at the same time.
The computer only looks for break through attempts and not for capture attempts, and therefore it didn't spot the white attack.
The matter of repetition and who did it has been a difficult subject always. Unhandyandy suggests in the large forum to simply let repetitions be black wins.

10-04/07:18 Hagbard: - And I believe that some of Ytreza's thought experiments last year also aimed at demonstrating, that it definitely cannot always be decided for a repetition, who is the "aggressor". This is a problem for a rule phrasing which presumes, that an aggressor is always known.

10-04/07:20 Hagbard: - The origin of the current rule phrasing is, by the way, the Foteviken Museum in Scone.

10-04/08:34 casshern: Please forgive my ignorance here, as I do not play Brandubh. Although move 15 is an attack on a white piece, white could move f3-g3 or king to d3. Not sure if either of these moves is strong or if white is simply in a losing position. The move repetition clearly starts on move 16 when white moves f3-f2 and black has to protect the piece while still trying blocking the corner (reacting to the aggressor).

I have gone into more details of my opinion on this matter in the large forum. If anyone would like to read and respond with their thoughts.

10-04/14:30 Adam: The repetition topic seems to be a bit spread around in the forum. I’ve had a read, and unless someone can prove that black can't force a repetition situation, the rule ought to apply to both. The idea that the game is already asymmetrical is not a good argument for the rule to only apply to white. On the contrary. I enjoyed the situation Ytreza proposes. Black needs to do something else when the cycle completes a second time. To clarify, I propose rule 8 read this: ’Perpetual repetitions are forbidden: the player who starts a repeating sequence of moves must find an alternative move to prevent a third cycle beginning, or else lose the game.’

It seems that the only reason this is a problem, is figuring out how to get the computer to recognise the repetition correctly. So no reason to change the rule. A human umpire can figure out cases of dispute when they crop up.

10-04/21:09 casshern: What we are talking about is the very essence of the game. The set up with black on the edge and white in the center, makes it so that white has to get past black. The point is that the king cannot do it on his own. He needs the help of his men to create and advantage to escape. Miller’s Guillotine proves that this can be done. And that is the point here. The responsibility of creating an advantage and not falling into a repetition falls on white. Be definition, black strategy is to repetitively and perpetually block white from getting past.

An issue I can see, is if black were to create an advantage for the win and either intentionally didn’t got for the win or lacked the skill to complete the win. And as the rules state, draw forts are a lose for white. So a repetition caused by a draw fort is still a lose for white. White should just resign or continue to try to create an advantage for a win than to create a repetitive position.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Mon Oct 05, 2020 2:52 pm

This discussion on white's and black's roles in repetitions called for an investigation of the games archive.

All games in the archive (all except for the new non-tafl types Saami Sahkku, Daabloe, Cuhkka and Hex) were searched through. 27,550 games all the way back to start of archiving Dec. 23rd 2010.

Not a single case of white winning a repetition was found.
Whereas 142 cases of black winning a repetion were found.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Wed Oct 07, 2020 9:14 am

Copied from the small forum:

10-06/22:24 Jrton80: I too remember Ytreza's thought experiment and I never bought into it. First, there were just too many logical gaps in what I read that I could not take it seriously. Second, I still have yet to see proof it is possible. I am not playing that phone app anymore since it calls Black repetition (well, that and the screen it too small for my old eyes), and no matter how many times I go through the the game, I am still not seeing it as black doing the repetition.

In My opinion, Black forcing a repetition is not a real thing. It has been a while now and I still have seen nothing to change my mind.

10-07/11:04 Adam: @Casshern, this is an argument that holds water, the very essence of the game. In chess the repetition rule (which incidentanly is based on position, not movement, so much for my previous comment!) results in a draw. In Tafl it is a loss, due to this essence of the game. @Jrton80, I agree, the Ytreza example was as full of holes as an old Viking's shield. But I enjoyed the idea of a longer looping sequence. I agree that noone has proved that black can force a repetition. However noone has proved that they can't either, which is rather the point of all this. I have decided to set myself the challenge of creating a board situation where black could force a draw by repetition. I doubt I can manage it. In the mean time, here is a suggestion: We could ammend the rule as has been suggested to a white loss, and ammend it further should a case of black forcing a repetition come up in game play. But wait!! That is precisely what has just happened (ChaolAibhne vs mmgaari), IF we agree that repetition is based on position as in chess, and NOT on a cycle of moves as I suggested below. IF repetition is based on position, then black CAN force a draw by repetition. Something an extremely cunning white player (mmgaari) could take advantage of if black is not paying attention. This is all very academic, but I see in Chess federation rules that the whole repetition business requires claims, checking of move history and confirmation by witnesses. Its a tricky business. It seems to me the onus in Chess is very much on the other player NOT to get themselves into a permanent checking position. But then, my eyes too are old and bent...

10-07/11:10 Adam: @myself, mind you, that would mean the advantage was still to white...so the repetition should be counted as a white loss. You guys are geniuses.

unhandyandy
Posts: 39
Joined: Fri Dec 28, 2018 6:39 pm

Re: Brandubh 7x7

Post by unhandyandy » Thu Oct 08, 2020 11:01 pm

Uh, what's the "small forum"?

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Thu Oct 08, 2020 11:27 pm

unhandyandy wrote:
Thu Oct 08, 2020 11:01 pm
Uh, what's the "small forum"?
The chat forum on
http://aagenielsen.dk/hnefatafl_online.php

Post Reply